etrk · HackerTrans

It's very interesting, especially if you grew up playing id games, and it's also an easy read. Go for it!

Regardless of road quality, weather conditions are unpredictable.

When the rain or snow or haboob kicks in, I guess your car will pull over to the side of the road and you'll need to drive manually until conditions improve. We'll need to keep the steering wheel around for the foreseeable future.

etrk

6 anni fadiscuss

Yeah, unless Leibnitz was busy making some tacos.

etrk

6 anni fadiscuss

In an alternate reality, Newton did five pull-ups instead of inventing calculus.

etrk

6 anni fadiscuss

Coyote sightings were a regular occurrence on my NextDoor feed long before the quarantine.

etrk

6 anni fadiscuss

"if its 1 dimensional then its signal processing"

This is clearly not the case, since image signal processing, 2D Fourier transforms, etc. are alive and well.

etrk

6 anni fadiscuss

It's asking for my email to check if my census data has been leaked? Email isn't even shown as a field in the leaked records. Is this site just harvesting emails?

etrk

6 anni fadiscuss

It's a recommendation by the FDA in the US, not an enforced regulation.

etrk

6 anni fadiscuss

Ah, I see what you mean. Yes, if you're not dealing with approximately bandlimited and sampled signals, then this wouldn't apply. The article is about embedded devices processing sensor data (microphones, motion/light sensors, accelerometers, etc.), and in those cases the signal of interest will often be bandlimited.

etrk

6 anni fadiscuss

> That is, signal processing had Nyquist's rates. And typically knows there is an underlying signal. Does ml have either?

What does this question mean? Every band-limited signal has a Nyquist rate. Most signals of interest are well-contained within some finite bandwidth (e.g., human voice). Sampling above this rate will get you very little.

If you're building an ML model to process a certain class of sampled signal and you know, for example, 99% of the signal energy falls within a certain frequency range, that should guide your choice of sample rate. If you're sampling at too high a rate, your input layers may have far more parameters than are needed or useful.

Whether or not a given ML input actually contains a signal of interest doesn't seem relevant to how you sample and preprocess the signal.

etrk

6 anni fadiscuss

What is a "typical person"? As the article states, most of these people are 70+ years old.

It would be interesting to know the prevalence of these illnesses among the cohort of 70+ year old Italians, but that would require more searching/translating than I care to do now. But feel free have at it.

etrk

6 anni fadiscuss

See Table 1 in the original study: https://www.epicentro.iss.it/coronavirus/bollettino/Report-C...

  Ipertensione arteriosa (high blood pressure) 76.1%
  Diabete mellito (diabetes) 35.5%
  Cardiopatia ischemica (heart disease) 33%
  Fibrillazione atriale (atrial fibrillation) 24.5%
  Cancro attivo negli ultimi 5 anni (cancer in past 5 years) 20.3%
  Insufficienza renale cronica (renal failure) 18%
  BPCO (COPD) 13.2%
  Ictus (stroke) 9.6%
  Demenza (dementia) 6.8%
  Epatopatia cronica (chronic liver disease) 3.1%

Note that the most of the people in this study were 70+ years old, and I have no idea what the typical prevalence of these diseases is among 70+ year old Italians.

etrk

6 anni fadiscuss

If you're working for a company where it's seen as "heroic" or "passionate" to come to work sick, you're in a toxic environment.

I've had a couple co-workers who perhaps see themselves as heroic for coming to work sick, but everyone else resents them and wishes they would just stay home.

etrk

6 anni fadiscuss

Real geniuses walk and drink their coffee at the same time.

etrk

6 anni fadiscuss

> "a problem that could not be solved"

The solution is counter-intuitive, but the problem can be solved with some elementary probability theory.

etrk

6 anni fadiscuss

I interviewed at some AI companies a year or two back. They all had teams of people dedicated to support each client: to clean their data, train their models, integrate the domain-specific requirements, customize UIs, etc. They sold themselves as the next AI-powered mega-unicorns, but they were more like boutique consultancies with no obvious path to scale up.

etrk

6 anni fadiscuss

> releasing practically their entire catalogue of movies on Netflix

Except in the US, Canada, and Japan.

etrk

6 anni fadiscuss

Citations tend to be a better indicator of impact. There are some interesting analysis here [1]. MIT has the most citations, and they've also published the most papers by far. If considering citations per paper, MIT falls in the middle of the pack. Toronto is an outlier with a very high number of citations per paper.

NeurIPs is a big conference covering lots of topics, though. All of this says relatively little about MIT's true impact on deep learning in the last eight years.

[1] https://www.microsoft.com/en-us/research/project/academic/ar...

etrk

6 anni fadiscuss

I’d heard that this is, to some extent, impossible, because everybody subvocalizes when they read. Wikipedia supports this, but I guess the goal is to minimize subvocalization rather than to eliminate it completely?

“Micro-muscle tests suggest that full and permanent elimination of subvocalizing is impossible.”

I know this is true for me. If I press the tongue to the roof of my mouth while reading, I can’t stop the muscles from moving very slightly as I read.

etrk

6 anni fadiscuss

I remember asking my French teacher if she ever “thought” in French, and she had no idea what I meant. She apparently had no internal monologue.

I assumed there was something with me for “thinking” in words. I tried to train myself to think more abstractly, and maybe it worked to some extent, but I’m not sure this was for the better.

comments